Baidu’s DuMate said it completed a core engine upgrade that reduced token consumption during task execution by 75%, with users’ points consumption also falling by 75%.
According to Jin10, DuMate said the reduction was achieved through its Harness engine and ongoing engineering-level tuning, while maintaining the agent’s intelligence capabilities and task execution performance.
DuMate said this marked the first time a domestic general-purpose intelligent agent product has delivered such a large drop in task-related consumption through the Harness engine and engineering optimization.
AI TRENDS | Baidu’s DuMate Cuts Token Use 75% After Core Engine Upgrade
2026-06-15 03:12:13
Disclaimer:
1. The information provided does not constitute investment advice. Investors should make independent decisions and bear all risks themselves.
2. The copyright of this content belongs to the original author. The views expressed herein are solely those of the author and do not represent the stance or position of this website.
Previous article：
Natixis’s García-Herrero: Strait of Hormuz Deal Eases Inflation Fears, Gives Fed More FlexibilityNext article：
汇丰、恒生部分网上服务今早一度受影响
